NSF-SNSF: ULTRA: Ubiquitous Large InTelligent ArRAys

毫米波（mm-波）和太赫兹（THz）频率的大量可用带宽正在为下一代通信和传感系统赢得越来越多的兴趣。在这些频率下，多天线收发器（TRX）阵列是克服高路径损耗和提高信噪比的关键，这对保持稳健的通信至关重要。与单天线系统不同，设计大型多天线阵列由于（i）失配和非理想性以及（ii）大量控制参数而带来重大挑战。现有的阵列主要依赖于查找表（LUT）和离线补偿。然而，这样的静态解决方案既不能捕获所有可能的操作模式，也不能提供对传播条件和通信/感测任务的实时自适应性。这项与瑞士苏黎世联邦理工学院研究人员的合作研究项目将产生具有高度可重构的140 GHz前端的新型智能大规模多天线阵列架构。与当今对阵列中的不平衡和失配视而不见或仅进行静态校正的系统不同，采用检测器和自适应阵列校准/波束形成将产生自适应阵列，从而随着条件的变化而表现得更好，并降低校准大型阵列的成本。该研究将直接影响超5G和6 G通信、亚毫米波雷达和下一代卫星/星间链路的应用。为了传播研究成果，该项目将提供阵列模型，阵列校准/波束形成代码和阵列演示测量的开放访问。ULTRA（无处不在的大型智能阵列）项目旨在通过综合考虑毫米波电子学，天线阵列，数字基带处理和校准算法的整体方法来解决大型阵列设计挑战。主要目标是设计D波段140 GHz的大型可扩展阵列架构，具有智能校准和自适应现场波束形成。工作分为几个方面，包括大型可扩展阵列中非理想性的分析和行为建模，设计可重新配置的D波段前端作为性能调谐“旋钮”，设计非侵入性D波段前端性能“传感器”，盲阵列校准和波束形成算法，以及ULTRA 140-GHz阵列系统集成和演示。两个波束独立的非理想性（例如，相位/增益/功率失配）和波束相关的非理想性（例如，天线耦合/串扰）将被建模。对于阵列天线负载补偿，将研究可重构负载调制平衡放大器发射机和具有非福斯特终端的Marchand巴伦接收机。D波段正交相位/增益调谐块也将探讨。将使用非侵入式现场有功功率和阻抗检测器检测天线失配/耦合（波束相关非理想性）。发射器/接收器（TX/RX）环回将测量阵列元件中的增益/相位/功率失配（波束无关非理想性）。将研究用于盲、动态阵列校准和残余损伤补偿的信号处理和机器学习算法。将在现场可编程门阵列（FPGA）原型板上实施数字波束成形方法，使其关键参数适应瞬时信道条件和校准/波束成形算法，以与D波段TRX阵列接口，用于ULTRA系统的离线/在线闭环操作。这个合作的美国-瑞士项目由美国国家科学基金会（NSF）和瑞士国家科学基金会（SNSF）资助，其中NSF资助美国研究人员，SNSF资助瑞士的合作伙伴。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The massive amount of available bandwidth at millimeter-wave (mm-wave) and terahertz (THz) frequencies is gaining increasing interest for next-generation communications and sensing systems. At these frequencies, multi-antenna transceiver (TRX) arrays are key in overcoming the high path-loss and boosting the signal-to-noise ratio, critical for maintaining robust communications. Unlike single-antenna systems, designing large multi-antenna arrays entails significant challenges due to (i) mismatches and non-idealities, and (ii) the large number of control parameters. Existing arrays mostly rely on look-up-tables (LUTs) and offline compensation. Such static solutions, however, can neither capture all possible operation modes nor provide real-time adaptivity to the propagation conditions and communications/sensing tasks. This collaborative research project with Swiss researchers at ETH Zürich will result in novel intelligent large-scale multi-antenna array architectures with highly reconfigurable 140 GHz frontends. Unlike today's systems that are blind to imbalances and mismatches in the array, or only make static corrections, employing detectors and adaptive array calibration/beamforming will result in an adaptive array, thus performing better as conditions evolve, and lowering the cost to calibrate a large array. The research will directly impact applications in beyond-5G and 6G communications, sub-mm-wave radar, and next-generation satellite/intersatellite links. To disseminate research results, the project will provide open access to the array models, array calibration/beamforming codes, and array demonstration measurements.The ULTRA (Ubiquitous Large inTelligent arRAys) project aims at addressing large-array design challenges with a holistic approach that jointly considers mm-wave electronics, antenna arrays, digital baseband processing, and calibration algorithms. The main objective is to design large and scalable array architectures at D-band 140GHz with intelligent calibration as well as adaptive in-field beamforming. The work is divided in several thrusts, including Analytical and Behavior Modeling of Non-idealities in Large Scalable Arrays, Designing Reconfigurable D-Band Frontends as Performance Tuning "Knobs", Designing Non-Intrusive D-Band Frontends Performance "Sensors", Blind Array Calibration and Beamforming Algorithms, and ULTRA 140-GHz Array System Integration and Demonstration. Both beam-independent non-idealities (e.g., phase/gain/power mismatches) and beam-dependent non-idealities (e.g., antenna coupling/crosstalk) will be modeled. For array antenna load compensation, a reconfigurable load-modulated-balanced-amplifier transmitter and a Marchand balun receiver with non-foster terminations will be investigated. D-band orthogonal phase/gain tuning blocks will also be explored. Non-intrusive in-situ real-power and impedance detectors will be used to detect antenna mismatches/coupling (beam-dependent non-idealities). Transmitter/receiver (TX/RX) loopbacks will measure gain/phase/power mismatches (beam-independent non-idealities) in array elements. Signal processing and machine learning algorithms for blind, on-the-fly array calibration and residual impairment compensation will be pursued. Digital beamforming methods that adapt their key parameters to the instantaneous channel conditions and calibration/beamforming algorithms will be implemented on field-programmable gate array (FPGA) prototyping boards to interface with the D-band TRX array for offline/online close-loop operation of the ULTRA system. This collaborative U.S.-Swiss project is supported by the U.S. National Science Foundation (NSF) and the Swiss National Science Foundation (SNSF), where NSF funds the U.S. investigator and SNSF funds the partners in Switzerland.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.